We have coped with rule generation from tables with discrete attribute values and extended the Apriori algorithm to the DIS-Apriori algorithm and the NIS-Apriori algorithm. Two algorithms use table data characteristics, and the NIS-Apriori generates rules from tables with uncertainty. In this paper, we handle tables with continuous attribute values. We usually employ continuous data discretization, and we often had such a property that the different objects came to have the same attribute values. We define a granulated table with frequency by discretization and adjust the above two algorithms to granulated tables due to this property. The adjusted algorithms toward big data analysis improved the performance of rule generation. The obtained rules are also applied to rule-based reasoning, which gives one solution to the black-box problem in AI.